HSP90 is a cellular chaperone that stabilizes several signal transduction networks important to cancer cells. Several years ago, we discovered that members of the benzoquinone ansamycin class of antibiotics, including herbimycin A and geldanamycin (GA) bound to HSP90 and disrupted its function. Multiple signal transduction proteins interact with these charperones, including the kinases src, erbB2, c-raf-1, Akt, Kit, Met, Bcr-Abl, the transcription factor HIF-1alpha, and mutated (but not wild type) p53. A general consequence of pharmacologic disruption of the chaperone/signal protein complex is the resultant marked instability and incorrect subcellular localization of the signalling protein. The instability is due to stimulation of targeted degradation of the signalling protein by the 26S proteasome proteolytic complex following chaperone dissociation. We have observed that geldanamycin reverses beta-catenin tyrosine phosphorylation in melanoma cells, probably due to the rapid loss of erbB2 from these cells. In untreated cells, erbB2 and beta-catenin can be readily co-precipitated. Loss of beta-catenin tyrosine phosphorylation leads to an increased association with E-cadherin and decreased cell motility in vitro. This is the first indication that modulation of the tyrosine phosphorylation status of beta catenin in melanoma cells is associated with decreased motility. The fact that beta-catenin tyrosine phosphorylation seems to be mediated, in 3/3 melanoma cell lines examined, by erbB2 - a geldanamycin-sensitive tyrosine kinase - suggests that geldanamycin treatment may be anti-metastatic. This hypothesis is currently being tested in an in vivo metastasis model. In collaboration with Brian Blagg of the Univ. of Kansas, we have identified a series of novobiocin derivatives that demonstrate improved binding affinity and anti-Hsp90 activity compared to the parental compound, and we have demonstrated the ability of several of these derivatives to deplete Hsp90 client proteins in tumor cells. The ErbB family of receptor tyrosine kinases contains four members. We have found that ErbB2, the only ligandless member of the family, is one of the most sensitive geldanamycin substrates. In contrast, mature ErbB1 is much less sensitive to geldanamycin than ErbB2. However, ErbB1 mutants found in NSCLC are highly dependent on Hsp90 and sensitive to its pharmacologic inhibition. In collaboration with Yong Lee of the Center for Information Technology, NIH, we have identified a short loop in the kinase domain of ErbB2 whose hydrophobicity and charge determine Hsp90 binding. We have shown that Hsp90 client kinases and non-client kinases differ in the charge distribution and hydrophobicity of this small region of the kinase domain. We recently identified the novel E3 ubiquitin ligase Chip as being recruited to ErbB2-associated chaperone complexes in the presence of geldanamycin. Chip mediates geldanamycin induced ErbB2 ubiquitination, which is necessary for, and precedes, its degradation by the proteasome. We have recently identified the site within the kinase domain of ErbB2 at which Hsp90 binds and we have proposed a model to explain remodeling of ErbB2-associated chaperone complexes in the presence of geldanamycin and other Hsp90 inhibitors. We have demonstrated that combination of low doses of geldanamycin and a proteasome inhibitor currently in clinical trial increases the toxicity toward tumor cells compared to that observed with each agent alone. We showed that this property was unique to tumor cells, in that the combination was not toxic to non-transformed cells at the concentrations tested. Further, we proposed a model to explain these results that invokes proteasome overload and deposition in the cell of insoluble proteins, leading to initiation of apoptosis. These data have led to initiation of a phase I combination clinical trial of an Hsp90 inhibitor and a proteasome inhibitor, in collaboration with investigators at the Mayo Clinic. We demonstrated that the trans-membrane endoplasmic reticulum signaling proteins, IRE1 and PERK, are Hsp90 client proteins and are thus sensitive to Hsp90 inhibitors. These data were the first to link Hsp90 to proper function of the unfolded protein response the stress response as it occurs in the endoplasmic reticulum. The data further underline the crucial role of Hsp90 in allowing cells to survive stressful stimuli. We recently demonstrated that HIF-1alpha, a transcription factor whose expression is upregulated by hypoxia and loss of VHL, is an Hsp90 client protein and is sensitive to Hsp90 inhibition. These data were the first to identify an oxygen- and VHL-independent pathway regulating HIF stability a pathway that is amenable to pharmacologic manipulation using Hsp90 inhibitors. These data have led to initiation, in the Urologic Oncology Branch, of a phase II clinical trial of benzoquinone ansamycins in clear cell kidney cancer (lacking VHL and expressing HIF-1alpha at constitutively high levels). We identified kinase-mutated KIT protein to be sensitive to Hsp90 inhibition. Kinase-mutated KIT is resistant to imatinib and other KIT inhibitors currently in the clinic. Kinase-mutated KIT is characteristic of mastocytosis and mast cell leukemia. Based on our findings, a phase II study of an Hsp90 inhibitor to treat mastocytosis has been initiated as an NCI/NIAID/Mayo Clinic collaborative trial. We demonstrated that MET and mutated MET protein, characteristic of papillary renal cell cancer, is sensitive to Hsp90 inhibition. Based on these findings, a phase II clinical trial of an Hsp90 inhibitor in papillary renal cell cancer is being initiated in the Urologic Oncology Branch. We identified Hsp90 on the surface of metastatic fibrosarcoma cells and demonstrated that specific inhibition of surface Hsp90 (using cell impermeable geldanamycin derivatives) significantly inhibits cell invasiveness. Further, we showed that inhibition of surface of Hsp90 prevented the maturation of MMP2, a metalloproteinase that has been implicated in fostering invastion and metastasis. Thus, we are now examining the model that surface Hsp90 may be a specific target in metastatic cancer cells. We have gone on to show, in a bladder cancer model system, that cell-impermeable Hsp90 inhibitors have dramatic inhibitory effects on cell migration and invasiveness, while having minimal growth inhibitory activity. We have extended these studies to include bladder, breast, prostate, and melanoma cancer cell lines with similar results. During the past year we have also shown that the combination of Hsp90 inhibition with proteasome inhibition is especially toxic to highly secretory tumor cells (while sparing normal cells). We have provided evidence that this drug combination disrupts the endoplasmic reticulum and therefore hypothesize that the endoplasmic reticulum is a viable target in cancer cells. We have also shown that Hsp90 inhibition is a potent radiosensitizer, in a cervical cancer model, both in vitro and in vivo. Lastly, we have provided evidence that geldanamycin cannot bind to Hsp90 without first being isomerized by the chaperone. This may provide binding specificity for activated forms of Hsp90, the state of the chaperone which is characteristic of cancer cells, but not normal cells. We speculate, therefore, that making use of this property may permit us to synthesize more potent, tumor Hsp90-prefering Hsp90 inhibitors